Dynamic response of materials and structures
- Predicting the dynamic response of materials and structures requires:
- Appropriate numerical tools for the job - coupling of solvers where a single tool is not suitable;
- Understanding the key governing rules of physics and ensuring the model assumptions are appropriate;
- Understanding sources of uncertainty (e.g. geometric, material, load, BCs, manufacturing process,…)
- Design optimisation / robustness;
- Experimental testing for material/component characterisation and model verification; and
- Stochastic design exploration.
A single deterministic numerical prediction is only part of the problem, particularly when dealing with complex, real world engineering problems. Here, an understanding of the sources of uncertainty (geometric, loading, material, manufacturing processes, etc.) and prediction of the damage / progressive failure is required to understand the behavioural response.
The accuracy of numerical models for depicting the behaviour of composite and metallic materials in simulations of continuum is one of the critical factors governing the overall accuracy and applicability of the results. In order to model dynamic events accurately the models should be able to account for all the important physical processes characterising material behaviour during a) the manufacturing process (e.g. curing, casting, deep drawing) including prediction of residual stresses and defects in composites and in doing so provide an accurate description of initial state of material for analyses; and b) dynamic events (car and aircraft crashes, bird strike, ice and debris impacts, earthquakes). This is particularly challenging where material anisotropy, large deformations, high strain rates, thermal effects, presence of shockwaves, damped vibrations or progressive damage and failure have to be considered. Furthermore, material softening (consequence of damage) causes additional numerical challenges.
To model progressive damage and failure, in addition to an accurate material model, appropriate spatial discretisation techniques are required. The two main methods used are the smoothed particle hydrodynamics (SPH) method and the finite element (FE) method. SPH and other mesh-free methods inherently allow for realistic modelling of initiation and propagation of damage (e.g. micro-cracks, voids) in arbitrary directions, which are not biased by a mesh structure and without the need for re-meshing. This makes the SPH method well suited for modelling local processes in and around the impact zone and damage localisation zones. On the other hand, global structural response is currently better modelled using the FE method. In order to combine the strengths of both the SPH and FE methods and enable multi-scale modelling of damage and fracture, a coupled FE-SPH method is under development. This novel approach enables the placement of cohesive interfaces at arbitrary positions (between particles) at the onset of free surface opening. Further, dynamic impedance matching between elements and particles will be ensured in order to accurately model the propagation of shock and stress waves and to prevent spurious wave reflections or high artificial stiffness.